Low reverse current LDO

[Unixon]

Hi,

I'm looking for a 5V->3V/3.3V linear regulator with low leakage current (1uA and less) in reverse powered configuration.

There's an USB powered device with optional battery backup.
During normal operation 3V bus is generated from 5V bus, and everything is OK.
However, when USB power is missing, neither LDO itself nor 5V bus powered circuit must suck the battery.
A diode at LDO input solves one half of the problem, but the other half with 3V-to-GND leakage remains.

Other requirements:
Load current could be as low as 10mA, however, since widely available LDOs provide at least 50-100mA and above it's likely any one of them will fit.
Output short-circuit protection, this thing must be unkillable. Small SMD package like SOT89 or SOT23, maybe some DFN or alike.

[Berni]

You will need a diode on the output of your LDO to fix this. To combat the voltage drop you can use a silicon diode that has a more predictable ~0.5V drop and raise the regulator voltage to compensate.

You can also add some circuitry to the FB pin of a regulator to make it have a adjustable current limit.

[exmadscientist]

Many parts have reverse current protection circuits. It seems likely you can find one that meets your specification (I'd start at onsemi personally), but if not the diode trick will help. With that it can be easier to maintain an accurate output voltage if you use the body diode of a low threshold n-channel MOSFET. You can then play with the gate to knock out the diode entirely when you don't want it in circuit.

[Unixon]

You will need a diode on the output of your LDO to fix this. To combat the voltage drop you can use a silicon diode that has a more predictable ~0.5V drop and raise the regulator voltage to compensate.

There's a little quirk to this solution
I thought of getting an adjustable regulator, but to offset the output diode voltage drop it must sense voltage after the diode with a voltage divider that is a current sink I'm trying to get away from.
Feedback circuit must be significantly lower impedance than the feedback pin itself, so I don't know if there's enough gap in impedance to make feedback leakage current low enough.
Also at least one of diodes (input) must be a schottky diode, otherwise total voltage drop may be out of range.

You can also add some circuitry to the FB pin of a regulator to make it have a adjustable current limit.

Unfortunately, almost no free space on PCB for these tricks Just relying on built-in overcurrent and thermal protections.

[wraper]

Use MOSFET based reverse current protection.

[Unixon]

Use MOSFET based reverse current protection.

Hmm... yeah, really... maybe I do have enough place for one or two SOT-23 MOSFETs.

[wraper]

You can also use a complementary pair in single package.

[wraper]

On second thought, there is an issue with my circuit. It protects from reverse current only if voltage on output is applied after input voltage was already removed. Otherwise Q2 will remain open. To overcome this flaw, gate of Q1 must be connected somewhere where voltage disappears when USB is disconnected even if output voltage is still present. For example USB VBUS with a Schottky diode added between VBUS and input of LDO, since LDO will pass current backwards through parasitic diode of its pass MOSFET.

[Warhawk]

Hi,

I'm looking for a 5V->3V/3.3V linear regulator with low leakage current (1uA and less) in reverse powered configuration.

There's an USB powered device with optional battery backup.
During normal operation 3V bus is generated from 5V bus, and everything is OK.
However, when USB power is missing, neither LDO itself nor 5V bus powered circuit must suck the battery.
A diode at LDO input solves one half of the problem, but the other half with 3V-to-GND leakage remains.

Other requirements:
Load current could be as low as 10mA, however, since widely available LDOs provide at least 50-100mA and above it's likely any one of them will fit.
Output short-circuit protection, this thing must be unkillable. Small SMD package like SOT89 or SOT23, maybe some DFN or alike.

What about going to ti.com -> products -> power management -> Linear & low-dropout (LDO) regulators -> view all products. From the filter on the left hand side -> Features -> Reverse polarity protection.
Then sort it according to the price from cheapest. What about TPS706? OC, OT, reverse current protection (100nA). Available on Mouser, Digikey, TI or in my drawer. 

[Berni]

There's a little quirk to this solution
I thought of getting an adjustable regulator, but to offset the output diode voltage drop it must sense voltage after the diode with a voltage divider that is a current sink I'm trying to get away from.
Feedback circuit must be significantly lower impedance than the feedback pin itself, so I don't know if there's enough gap in impedance to make feedback leakage current low enough.
Also at least one of diodes (input) must be a schottky diode, otherwise total voltage drop may be out of range.

That's why you need to use a diode with a consistent voltage drop between the regulator output and load, since the divider has to be before the diode. That way you can compensate for the voltage drop on the diode by setting the regulator slightly higher.

When it comes to microAmps of reverse leakage you likely have no other choice.

On second thought, there is an issue with my circuit. It protects from reverse current only if voltage on output is applied after input voltage was already removed. Otherwise Q2 will remain open. To overcome this flaw, gate of Q1 must be connected somewhere where voltage disappears when USB is disconnected even if output voltage is still present. For example USB VBUS with a Schottky diode added between VBUS and input of LDO, since LDO will pass current backwards through parasitic diode of its pass MOSFET.

Yep this is actually reverse polarity protection, not reverse current protection. The FET turns on whenever there is positive voltage present. In order to have it work for reverse current you actually need a ideal diode controller chip that can detect the minute negative voltage across the FET in reverse flow to turn it off.

However you could likely make it work if you connected the sensing line of the N FET to the voltage input before the regulator.

What about going to ti.com -> products -> power management -> Linear & low-dropout (LDO) regulators -> view all products. From the filter on the left hand side -> Features -> Reverse polarity protection.
Then sort it according to the price from cheapest. What about TPS706? OC, OT, reverse current protection (100nA). Available on Mouser, Digikey, TI or in my drawer. 

That is an interesting chip. Usually reverse protection just means that you can hook up the input the wrong way around without smoke. Tho in the chip shortage i would advise against using specialized chips like that if the solution can be done with a regular standard jelly bean regulator and some extra parts around it.

[T3sl4co1l]

No need, there are LDOs with reverse rectification function.  Usually for automotive application, where the input reversal would have the same problem.

Like, have seen this before, https://www.onsemi.com/pdf/datasheet/ncv4274-d.pdf Not really clear if they suggest/approve of powered-output configuration but it's probably fine.  I would definitely get one to test (I haven't done it myself, at least, not sure, it's been a while).  This one doesn't seem to be well stocked (only the D/2PAKs are at DK right now) but there are others, also check Infineon, TI, etc.

Tim

[Warhawk]

Hupps, I meant reverse current protection (the filter).

What about going to ti.com -> products -> power management -> Linear & low-dropout (LDO) regulators -> view all products. From the filter on the left hand side -> Features -> Reverse polarity current protection.
...